search

rachitshah86 / guardd

Automatically exported from code.google.com/p/guardd
0 stars 0 forks source link

Jackknife method for error estimation (alternative to Monte Carlo) #17

Open GoogleCodeExporter opened 8 years ago

GoogleCodeExporter commented 8 years ago 
GUARDD currently uses the Monte Carlo (MC) method of error estimation. Many 
other CPMG RD analysis programs use the "jackknife" method instead / in 
addition. The jackknife method is a powerful alternative that iteratively 
re-fits the data using a different a subset of the entire dataset at each 
fitting iteration.

How to program
 (1) Add an option to specify the MODE of error estimation (edits in Session.m)

  (a) Session.m: Add parameter ERROR_ALGORITHM which is a string that will be set as either 'MONTECARLO' or 'JACKKNIFE'

  (b) Session.m: Allow the user to change this parameter in the Settings GUI window
    Add this parameter name and a description to param_info near the top of Session.m    

  (c) Session.m: Make sure that only allowable values of this new parameter can be provided by the user
    Edit Session.m/saveDataFromTable function

    Make new code to test the allowable options (it will look similar to MC_RANDOMIZE_MODE)

 (2) Use this new parameter to specify the mode of error estimation (FitResult.m)
  (a) FitResult.m/calculateErrors: Check which mode of error estimation should be used
    Check obj.parentGroup.parentSession.ERROR_ALGORITHM

  (b) Program the Jackknife procedure

  (c) Make sure that errors are properly stored in the same data structures that were used for the Monte Carlo method

Time required
 5-20 hrs

Difficulty
 Medium

References

Original sources
(1) Mosteller, F. & Tukey, J. (1968) Data analysis, including statistics. In: 
Lindzey G & Aronson E (Eds.) Handbook of Social Psychology, Vol. 2, 2nd edn. 
Addison-Wesley, Reading, MA, pp. 80-203

(2) Mosteller, F. & Tukey, J.W. (1977). Data Analysis and Regression: A Second 
Course in Statistics. Addison Wesley.

Reviews on the method
(3) A Leisurely Look at the Bootstrap, the Jackknife, and 
Cross-Validation.Detail Only Available By: Efron, Bradley; Gong, Gail. American 
Statistician, Feb83, Vol. 37 Issue 1, p36

(4) Efron, Bradley. The jackknife, the bootstrap, and other resampling plans. 
Philadelphia, Pa. : Society for Industrial and Applied Mathematics, 1982

Original issue reported on code.google.com by ian.klec...@gmail.com on 11 Sep 2011 at 9:43

GoogleCodeExporter commented 8 years ago 
Review
JOSEPH L. BALLOUN and A. BEN OUMLIL

JACKKNIFE: A general-purpose package for generating multivariate jackknife 
analyses

Behavior Research Methods, Instruments, & Computers
1986, 18 (1), 47-49

Original comment by ian.klec...@gmail.com on 11 Sep 2011 at 10:05